cluster annotation in single-cell (-nuclei) RNA-Seq or multiome experiments

Question

[gogeni5529]

I'm new to single-cell RNA Seq and nowworks on a data set, for which we don't really have a lot of marker genes (we want to try some new ones).

Is it possible to classify a cluster based on just one gene with distinct expression profile? How many genes does one need for cluster annotations. I have seen papers (e.g. here) where they used just one gene for that. Is this stable enough?

I was wondering if there are tools to try and identify (calculate?) gene markers based on expression profile or behaviour of specific genes. I know of SingleR, but this unfortunately didn't provide satisfying results in our case.

thanks

[biofalconch]

Usually tools use K-nearest neighbour methods to do the clustering. Many papers report only a single marker genes since no other gene is differentially expressed based on the FindMarkers function. Maybe before jumping to something else you could play around by switching the resolution in the FindClusters function.

[gogeni5529]

thanks biofalconch for the answer. I don't mind having just one gene. My question was more, if I can trust the results with just one gene. it will make my life a lot easier :-)